Re: [PATCH v3 1/4] dt-bindings: clock: mediatek,mtmips-sysc: Adapt compatible for MT7688 boards

[Krzysztof Kozlowski <krzk@kernel.org>]

On 24/06/2025 11:21, Ezra Buehler wrote:
>> On 24 Jun 2025, at 11:12, Krzysztof Kozlowski <krzk@kernel.org> wrote:
>>
>> On 24/06/2025 11:08, Ezra Buehler wrote:
>>>> On 20 Jun 2025, at 09:42, Krzysztof Kozlowski <krzk@kernel.org> wrote:
>>>>
>>>> On Thu, Jun 19, 2025 at 10:34:59PM GMT, Ezra Buehler wrote:
>>>>> From: Ezra Buehler <ezra.buehler@husqvarnagroup.com>
>>>>>
>>>>> As the MT7628 and MT7688 are identical in most respects, mt7628a.dtsi is
>>>>> used for both SoCs. To prevent "Kernel panic - not syncing: unable to
>>>>> get CPU clock, err=-2" and allow an MT7688-based board to boot, the
>>>>> following must be allowed:
>>>>>
>>>>>   compatible = "ralink,mt7628-sysc", "ralink,mt7688-sysc", "syscon";
>>>>>
>>>>> Signed-off-by: Ezra Buehler <ezra.buehler@husqvarnagroup.com>
>>>>> ---
>>>>> .../bindings/clock/mediatek,mtmips-sysc.yaml  | 27 ++++++++++---------
>>>>> 1 file changed, 15 insertions(+), 12 deletions(-)
>>>>>
>>>>> diff --git a/Documentation/devicetree/bindings/clock/mediatek,mtmips-sysc.yaml b/Documentation/devicetree/bindings/clock/mediatek,mtmips-sysc.yaml
>>>>> index 83c1803ffd16..3fabaa8acc10 100644
>>>>> --- a/Documentation/devicetree/bindings/clock/mediatek,mtmips-sysc.yaml
>>>>> +++ b/Documentation/devicetree/bindings/clock/mediatek,mtmips-sysc.yaml
>>>>> @@ -26,18 +26,21 @@ description: |
>>>>>
>>>>> properties:
>>>>>  compatible:
>>>>> -    items:
>>>>> -      - enum:
>>>>> -          - ralink,mt7620-sysc
>>>>> -          - ralink,mt7628-sysc
>>>>> -          - ralink,mt7688-sysc
>>>>
>>>> I do not understand why this is removed and commit msg explains nothing
>>>> about it. Re-add it back.
>>>
>>> OK, so you suggest we allow
>>>
>>> compatible = "ralink,mt7628-sysc", "ralink,mt7688-sysc", "syscon";
>>>
>>> and
>>>
>>> compatible = "ralink,mt7688-sysc", "syscon";
>>>
>>> I'll adapt my patch accordingly.
>>>
>>> Still, as AFAIK the MT7628 and MT7688 are identical in this regard,
>>
>> Standard rules apply expressed in writing bindings, some talks/guides
>> and all modern SoCs...
>>
>>
>>>
>>> compatible = "ralink,mt7688-sysc", "ralink,mt7628-sysc", "syscon";
>>>
>>> would technically be valid too. Could you elaborate why that is not a
>>> good idea? The MT7688 is basically a subset of the MT7628.
>>
>> You did not send such patch.
> 
> No, I did not. I am just trying to understand the topic better. Anyway,
> never mind, I will just do it as you suggested.


I cannot give you answer on hypothetical patch without seeing that
patch. Maybe you change or break the ABI? Maybe you affect users? Dunno.
I prefer to discuss the code, not scenarios.


Best regards,
Krzysztof